Therapeutic potential of targeting gene variants in schizophrenia

Schizophrenia is a debilitating, chronic syndrome affecting 1% of the population worldwide with compelling medical, economic and ethical reasons for improving existing treatments. Unfortunately, despite intensive efforts by the pharmaceutical industry, the current drug discovery effort for developing improved treatments for schizophrenia via a ‘one drug fits all’ blockbuster approach has stalled. Although the exact reasons are complex, the numerous scientific obstacles include diagnostic uncertainties, limitations of preclinical models in nonverbal animals, lack of knowledge about etiology and the complexity of this illness (reviewed in [1]).

In particular, the heterogeneity of schizophrenia presents a major challenge with many etiologic routes leading to the same complex end point, a phenomenon shared with the broad syndromes of headaches, epilepsy or cancer. Schizophrenia is one label for many separate disease processes; a lesson we have come to appreciate recently for nonfamilial sporadic autism [2], which may present with similar phenotypes caused by the deletions of many different genes. Still, the search for a common final pathway for schizophrenia is seductive, but may represent a misguided frame for studying the etiology. Nonetheless, important family studies have defined the risk to first-degree relatives of patients with schizophrenia as 8–10-times the risk in the general population. Longitudinal epidemiologic studies have convincingly shown multiple environmental risk factors for schizophrenia such as time of birth (winter or urban), in utero infections, malnutrition, obstetric complications and paternal age as 1.2–5-fold risk factors. From numerous genetic studies over the last decade and the first whole genome association studies, the genetic architecture of schizophrenia and complex traits in general appears to consist of numerous small magnitude genes as risk factors individually conferring a 1.1–1.5 odds ratio [3–5] and updated via the schizophrenia gene website [101]. Given this reality, drug development approaches may profit from accepting that there are a multitude of pathways to psychosis, and should create personalized approaches for some of these pathways. We envision patients will benefit from a combination or cocktail of drugs similar to those now used in refractory cases of other complex disorders such as cancer, diabetes and hypertension.

How can genetics contribute to this process? In general, the identification of genetic variants influencing schizophrenia can make contributions to four distinct domains:

•	Illumination of etiology
•	Refinement of the psychiatric classification system
•	Identification of high-risk individuals for early interventions
•	Discernment of pathways and target molecules for drug intervention

In addition, some stark realities need recognition. First, environmental risk factors as well as comorbid processes (e.g., cannabis abuse) play important roles in promoting and maintaining the psychotic process. If gene x environment interactions as well as comorbidities contribute to the biological vulnerability for a given developmental pathway, treatment development will have to consider these effects. Second, if schizophrenia is an illness that is both neurodevelopmental and results from rare variants, we will need to create an in-depth, systems-based approach to the understanding of the developmental changes associated with uncommon DNA variation. This systems approach might cluster a myriad of ‘genetic nicks and cuts’ into a handful of perhaps inter-related mechanisms whose net behavioral effects leads to schizophrenia. We provide two examples of genetic variants to further illustrate this view, namely a rare variant of a Mediator complex gene (MED12^12bp also named HOPA^12bp) and the Neuregulin 1 gene (NRG1).

The Mediator complex is a large (∼25 units) multiprotein assembly that integrates and conveys a wide variety of regulatory signals, including those of the Wnt, nuclear receptors and receptor tyrosine kinase (RTK) pathways, by physically linking transcriptional regulators to the RNA polymerase II transcriptional assembly [6,7]. MED12 is the second largest protein subunit in this transcriptional complex and has critical roles in each of the three above pathways [8]. In 1998, we reported the discovery of the MED12^12bp (HOPA^12bp) variant, which is a four amino acid insertional polymorphism in the portion of the gene (Opa domain) responsible for mediating Wnt pathway signaling. Over the past several years, we and others have shown that this relatively uncommon (1 in 60 X chromosomes in European populations) balanced polymorphism is a mild risk (∼1.5-fold) for a syndrome of positive symptom psychosis [9,10]. In addition, we have identified other allelic variants (e.g., HOPA^15bp) in the same gene domain that are associated with risk for illness.

Using a systems biology-based approach, we hope to capitalize on others’ work delineating MED12’s role in the Wnt pathway. The goal is to identify a potentially coherent pathway to psychosis for some patients. The Wnt proteins are a family of secreted signaling ligands that bind to Frizzled and lipoprotein receptor-related protein family receptors, regulating numerous fundamental processes in neural development. Disruption of Wnt 1, 3 and 5 alters dopaminergic growth and differentiation [11]. Intriguingly, ablation of MED12 has selective effects on neuronal and neural crest differentiation and, in particular, on dopaminergic neuronal growth and differentiation [12,13]. Since positive symptoms of psychosis are hypothesized to be a consequence of dopaminergic dysfunction and MED12 regulates Wnt pathway signaling, these studies suggest that genetic variation at points throughout the Wnt signaling pathway may define a distinct psychotic diathesis whose features could be studied by using the already identified MED12 sequence variation as an exemplar.

In fact, the literature is rich with studies indicating that alterations in Wnt signaling may contribute to psychosis [14–16]. Additional studies show antipsychotic treatment alters Wnt pathway signaling [17]. These findings suggest that genes which form hubs or nodes downstream of the Wnt pathway, such as glycogen synthase kinase-3, could be pharmacologically targeted [18]. However, to date, those efforts have not come to clinical fruition, and even if they do, the detection of selective beneficial effects of such ‘Wnt-specific’ agents in clinical trials may be hindered by the broad genetic heterogeneity present in schizophrenia. This is unfortunate because variants in this pathway may have distinct clinical profiles that may be more aptly addressed by some agents more than others.

While in principle segmenting the population of patients by pathway seems feasible, how would one cluster these subjects and develop specific agents for these subjects? A possible answer is the blending of basic science, systems biology and bioinformatics. For example, using the above MED12 example, one could speculate and imagine a two-stage approach for identifying a subpopulation for a targeted treatment. Using a simpler zebrafish model system and a forward mutagenesis strategy, one could determine the relationship between sequence variation in Wnt pathway genes on dopaminergic growth, differentiation and function, yielding multiple alleles and their effects. Then, allelic variation data from resequencing psychotic subjects’ information could potentially identify individuals with significant loss- or gain-of-function alleles in the Wnt pathway genes that mimic the effects on zebrafish dopaminergic function. Next, using the same zebrafish models, one could theoretically conduct preclinical testing of candidate drugs. Yet, zebrafish are not only mute, but also do not develop schizophrenia. However, it might eventually be possible to construct genetic models mimicking existing human polymorphisms in zebrafish. In summary, while the use of specially constructed zebrafish models won’t necessarily speed later phases of drug development, their use may facilitate the identification of subsyndrome specific agents.

A second illustrative example of the possible therapeutic use of a schizophrenia risk allele examines NRG1, a transmembrane EGF-domain containing ligand with several isoforms. NRG1 was initially defined as a 1.8–2.2 relative risk for illness by genetic association studies by DeCode and collaborators, with a risk haplotype near the 5´ end of the gene [19]. Recent studies have estimated the effect as a 1.5 relative risk factor with a risk haplotype in a noncoding, regulatory region (reviewed in [20]). NRG1 has many functions across development, affecting neuronal migration, synapse formation, neurotransmitter receptor regulation and even long-term potentiation. The mechanistic pathway of action includes the sequential steps of the release from a presynaptic neuron of the membrane bound or soluble form of the growth factor NRG1, binding of the NRG1 ligand to the heterodimeric ErbB3–ErbB4 receptor complex (v-erb-b2 erythroblastic leukemia viral oncogene [avian] ErbB; a receptor tyrosine kinase) and signaling via a downstream kinase cascade. The release of the transmembrane NRG1 ligand by metalloproteases may be one drugable target. Other potential targets for drug treatment include the downstream signaling pathways, including the autophosphorylation of the ErbB receptor on a cytoplasmic tail tyrosine, the subsequent recruitment of docking proteins and the activation of Ras–ERK/MAPK and PI3K–AKT pathways. While innumerable critical details regulating this NRG1–ErbB3/ErbB4 pathway across time and location in the brain remain to be characterized, these ligands and receptor tyrosine receptors provide a potentially rich set of drugable targets. Specifically, it may be possible to develop molecules to sequestrate and bind the ligand NRG1 or develop small molecule antagonists to inhibit binding to the ErbB3/ErbB4 receptor. Alternatively, given the intense work on RTK, one can imagine inhibitors of the ErbB3/ErbB4 receptor dimerization or tyrosine kinase activity, leading to reduced autophosphorylation. Finally, one could target the ErbB3 receptor levels controlled by the E3 ubiquitin ligase (neuregulin receptor degradation protein-1/ring finger protein 41), which ubiquinates the ErbB3/ErbB4 receptor, tagging the receptor for degradation. In any case, even if these drugs were available, the challenges remain to define the appropriate subpopulation to treat and determine the appropriate age of intervention for a neurodevelopmental disorder. Hence, finding a gene and turning it into a useful intervention requires a much fuller understanding of its pathophysiology across time and which of its pleiotropic functions are linked to vulnerability to illness.

Finally, the most feasible current therapeutic use of gene variants (and future whole genome single nucleotide polymorphisms associated with schizophrenia) is as biomarkers to identify those individuals at risk for schizophrenia. These high-risk individuals might be targeted for more vigilant monitoring and early interventions such as psychosocial therapies (e.g., cognitive remediation, vocational rehabilitation, family interventions and cognitive–behavioral therapy), drug therapy and other prevention strategies. While these biomarkers will be individually weak predictors, the aggregation of genetic risk factors combined with clinical history, family history, environmental risks, neuroimaging, neurocognitive testing and premorbid functioning should improve the odds of defining a high-risk target group for intervention and matching patients to treatments with more acceptable risk–benefit ratios.

In summary, the genetics of schizophrenia remains complicated, messy and unfinished, yet the field continues to mature with the ascent of well powered, large case-controlled whole genome association studies (e.g., the Genetic Assocation Information Network [GAIN] study [102]), forming a foundation for functionally determining the identity and subsequently the roles of small magnitude gene effects. Over the next decade, as these gene variants are identified and replicated and their interaction with environmental risks examined, these genes can be skillfully clustered into functional networks and pathways, as has been done already in yeast [21]. From this network and systems view, the hubs, nodes or switches suggest points of leverage within pathways for interventions. We continue to believe these improvements will be incremental and not revolutionary with few if any blockbuster drugs. However, as the etiology becomes clearer and the drugable targets arise, the field becomes more attractive for interventions. The hallmark of progress will be an understanding of the etiology of schizophrenia. We anticipate that this new knowledge through the study of genetic variants will allow more rational design of drugs, clarify the nosology of schizophrenia and target more appropriate subpopulations for intervention.

Acknowledgements

RA Philibert is supported by DA015789. The NIH and The University of Iowa have been granted a patent covering the detection of HOPA polymorphisms as a test for increased susceptibility to schizophrenia. RA Philibert is a potential royalty recipient. HK Gershenfeld is generously supported by NIH (RO1 MH067211) and a NARSAD Independent Investigator Award.